JPH11191716A - Simulation method for nonlinear amplifier having envelope memory effect - Google Patents

Abstract

PROBLEM TO BE SOLVED: To simulate a response signal of a nonlinear amplifier with a memory effect. SOLUTION: The method is provided with a block 31 that calculates an output signal (y) and a harmonic generating block 32 in parallel with the block 31 that calculates frequencies of the inter-modulation component to correct the output signal (y). In the simulation method for a response of an nonlinear amplifier, the nonlinearity of amplitude/phase shift conversion of an amplifier with constant amplitude at an input side is measured, its characteristics of different frequencies is measured, a frequency correction value for a direct transfer function is calculated to simulate an envelope memory effect, and amplitude modulation distortion is measured by modulating the input amplitude. A modulation transfer function to reproduce an amplitude of distortion of an output side is calculated depending on the input modulation amplitude, and the direct transfer function is corrected when the input amplitude is modulated so that expansion of the characteristics in a series of the direct transfer function is attained.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a simulation of a signal response of a nonlinear amplifier.
An object of the present invention is a system for simulating a response signal of a nonlinear amplifier having a memory effect. This type of system is used to simulate high-efficiency microwave amplification, especially AB, B or C class amplification, especially solid state current amplifiers (SSPA) and traveling wave tube amplifiers used for terrestrial or satellite broadcasting. (TW
It can be used for simulation of the response of T). This type of amplifying device has a response curve unique to nonlinearity at high frequency or high efficiency.

[0002]

2. Description of the Related Art FIG. 1 shows an example of an input / output response curve of a nonlinear amplifier ANL. The curve giving the output signal level g as a function of the input signal level x of the amplifier ANL usually bends when the input signal x has a large amplitude A due to the saturation phenomenon. When an amplifier is used with a non-constant gain as a function of the input signal level x, the amplifier is said to be operating in a non-linear mode or its amplification is simply referred to as non-linear amplification. Non-linear devices can be divided into devices without memory, devices without quasi-memory, and devices with memory.

[0003] Amplifiers without memory have high non-linearity in amplitude and low phase distortion. Therefore, by modifying the input / output response characteristics of the amplifier ANL without memory,
As shown in FIG. 1, a single curve g (x) can be obtained. It is possible to model and simulate the response of the memoryless amplifier ANL as a function of time t into a sinusoidal input signal x of frequency f ₀ , amplitude modulation A, phase modulation φ. This signal is represented as:

[0004]

(Equation 1)

A (t) represents the envelope of the input signal, which is defined by the limit value of the amplitude, at which the sine signal x is developed, and the envelope changes as a function of time. FIG. 3 shows a timing diagram of a signal x (t) having a constant envelope A. On the other hand, the output signal g of the amplifier without memory is expressed by the following equation (2).

[0006]

(Equation 2)

Since memoryless amplifiers have an instantaneous response, it is useful to provide these signals x and g with a complex envelope when discarding the reference at time t. The output signal x represented by equation (1) has a complex envelope X represented by equation (3) below:

[0008]

(Equation 3)

All useful information about the amplitude modulation A (t) and the phase modulation φ is recorded in this complex envelope X. Similarly, the output signal g represented by the equation (2) has a complex envelope G represented by the following equation (4):

[0010]

(Equation 4)

For the non-linear amplifier ANL without memory, it is shown that C (A) is a Chebyshev transform of the input / output response curve g (x). Therefore, the non-linear amplifier AN without memory
The response of the L to the modulated signal x can be modeled and simulated using only a single curve C (A), as exemplified by the solid line in FIG. Such a curve representing the output amplitude C as a function of the input amplitude A is called a non-linear curve in amplitude, and in short form is called an AM / AM curve.

FIG. 2 shows a non-linear amplifier ANLA with a memory.
A response specific to M is shown, and a hysteresis phenomenon caused by the memory effect is observed. It can be seen that the rising hysteresis curves m and m 'do not overlap each other when the amplitudes A and A' of the input signals are different. Since the time of the memory changes in relation to the change in the amplitude, the curve m
And m 'cannot overlap each other.

If the time of the memory of the amplifier ANLAM is negligible compared to the period of the amplitude variation A (t), the amplitude A is furthermore considered to be constant and the amplifier is called a quasi-memoryless amplifier. FIG. 4 shows a timing diagram of the output signal y of the amplifier without quasi-memory, in which the expression (1) to which the input signal x shown in the timing diagram of FIG.

[0014]

(Equation 5)

For a non-quasi-memoryless non-linear amplifier to which the signal x of equation (1) applies, the output signal y is given by:

[0016]

(Equation 6) Here, C (A) is the amplitude of the output signal y, and φ (A) is the phase shift (shift) of the output signal y depending on the amplitude A (t) of the input signal.

Therefore, at a certain instant t, the amplitude C (A) of the output signal y and the phase shift φ (A) depend only on the amplitude A of the input signal x at the moment t. Amplitude change A
It is possible to ignore (t) as a function of time and assume that the amplitude A is almost constant as seen in FIG. It is also useful to express the envelope of the signals x and y in a complex manner. The envelopes X and Y are expressed by the following equations (3) and (6), respectively.

[0018]

Equation 7 (Envelope A is considered constant over time)

[0019]

(Equation 8)

Thus, the response of a nonlinear amplifier without quasi-memory can be modeled and simulated based on knowledge of the following two characteristic curves: the output signal y amplitude as a function of the input signal amplitude A C (A) [Curve AM shown by a solid line in FIG.
/ AM (amplitude / ampli
curve) which represents the phase shift φ of the output signal y with respect to the input signal x as a function of the amplitude A of the signal x,
An amplitude / phase conversion curve, called AM / PM for short, is shown by a solid line in FIG. Similarly, curve C (A) is the norm of the complex Chabyshev transform of response characteristic y (x), and curve φ (A)
Is the argument of the complex transformation. The complex envelope Y of the output signal can also be written in two parts. That is, the real part and the imaginary part correspond to the in-phase component P and the quadrature component Q. The components P and Q are represented by the following equations (7 ') and (7 "), respectively.

[0021]

[Equation 9]

FIG. 7 shows a curve C in FIGS. 6 (a) and 6 (b).
An example of curves P (A) and Q (A) corresponding to (A) and φ (A) is shown. Although the two sets of curves are exactly equivalent, one commonly used model for simulating the response of known quasi-memoryless amplifiers is C
It is the characteristic of a pair of curves P (A) and Q (A) rather than (A) and φ (A).

According to known principles of nonlinear amplifier simulation, existing amplifiers are pre-characterized on a test bench. Prior characteristic measurement is performed using a signal having a specific frequency and amplitude in order to simulate a response to a signal having an arbitrary frequency and amplitude.

[0024] As shown in Figure 5, different amplitudes A ', A ", A'" signal having a single frequency f ₀ which is taking is to obtain the characteristics shown in FIG. 6 or FIG. 7, for example As applied to the tested amplifier. However, for an amplifier with a certain amount of memory, the characteristics are that the frequency of the signal to be amplified, f ₀ ⁻ and f ₀ and f ₀ ⁺
It has been confirmed that it changes considerably in response to

[0025]

A known system for the simulation of such amplifiers is described in US Pat.
B. Poza (H.B. Poza) has published the document "A Wide
band Data Link Computer S
immulation Model "," NAECON '
75 Record ", page 71. In the literature, the frequencies f ₀ ⁻ , f
Several pairs of curves AM / AM and AM for ₀ or f ₀ ⁺
It is proposed to plot / PM. FIG.
(A), FIG. 6 (b) shows three pairs of AM / AM and AM obtained at each of three frequencies f ₀ ⁻ , f ₀ and f ₀ ⁺ located in the effective band BU of the directional radio link amplifier. 4 shows a / PM curve.

H. B. Posa's simulator is AM
Only one pair of / AM and AM / PM curves is stored. For example storing a pair of curves obtained at the frequency f _0, other frequencies f ₀ ^-, to reconstruct the curve that is not stored corresponding to the frequency of f ₀ ⁺ or intermediate. Unstored curves are derived by simply moving the stored curves by the appropriate vector. The simulator is shown in Fig. 6.
(A) or the A axis of the movement vector and C so that the curve of the frequency f ₀ shown by the solid line in FIG. 6B is as close as possible to the curve of the frequency f ₀ ⁻ or f ₀ ⁺ shown by the dotted line.
Calculate the axis or φ axis components.

This type of simulation system is too approximate to simulate the distortion occurring at different frequencies in a quasi-memoryless amplifier. Another disadvantage of this type of system is that it cannot be used to simulate the response of an amplifier with memory.

FIG. A. M. Saleh (AA)
M. Another known simulation system based on the model of Saleh) is described in the document "Frequen.
cy-Independent and Freque
ncy-DependentModels of TW
TA Amplifiers ", November, 1
982, "IEEE Transactions on
Communication ", Volume Co
m-29, No. 11 shows the system described in page 1715. The calculation of the response to the input signal x of the non-quasi-memoryless non-linear amplifier is given by P (A,
f) and Q (A, f).
The component yp is in phase with the input signal x, and the component yq is in quadrature.

The Surrey simulation system uses the results of amplifier characteristic measurements at a plurality of frequencies f. That is, transfer functions P (A, f) and Q corresponding to each arm for calculating components yp and yq of output signal y.
Several curves P (A) and Q (A) are plotted at several frequencies f to calculate (A, f).
Since the phase shift is not introduced in the respective calculation branches, the transfer function P (A, f) or Q (A, f) respectively corresponds to the calculated value of the non-linear response without memory effect.

A. A. M. Surrey's document states that this model of the system can be applied to the amplification of single frequency signals, and speculation assumes that it can be applied to the simulation of any type of signal. Another disadvantage of this model is that when the amplitude of the signal changes rapidly with respect to the time constant of the memory, the distortion that appears at different frequencies cannot be reproduced by this system, so it must be applied to an amplifier with memory. Is not possible.

In the non-linear amplifier ANLAM with a memory, the time required for the memory is not negligible compared to the time when the amplitude A (t) changes, so that the memory effect is even greater. In such a case, more complex methods are needed to simulate the response of the memory-equipped amplifier ANLAM, such as a non-linear differential equation or the expansion of a characteristic curve into a function sequence.

A known and improved model is described in the document "Frequency-Dependent Nonl.
inner Quadrature Model fo
r TWT Amplifiers ", August
1984, "IEEE Transactions o
n Communication ", Volume C
om. 32, no. 8, page 982. T. Abuermaatti (MT Abuelm
a'atti). This model allowed the simulation of the response of the amplifier with memory using the evolution of the characteristic curve of the nonlinear amplifier in a sequence of Bessel functions.

FIG. T. Abuel Maatti represents the simulation system described above, which consists of two calculation branches, an in-phase component yp and a quadrature component yq of the output signal y, where each arm is a frequency correction factor. Each function J1 is weighted by α and the factor G _n (f), and N depends on the amplitude A of the input signal x.
The contribution of the Bessel function sequence J1 is calculated. The coefficient α _n and the factor G _n (f) are obtained by plotting the characteristic curves P (A) and Q of the nonlinear amplifier plotted at a plurality of test frequencies f of the amplifier.
It is calculated after setting several pairs of (A).

In FIG. 8, the weighting coefficients α _np and α _np of these first-order N-bessel functions, which are called J1 (nπA / D), are obtained by summing the two weighted sums.
_It can be seen that the curves P (A) and Q (A) as shown in FIG. 7 can be interpolated very accurately while adjusting _nq .

The Bessel function is a Chebyshev transform of the sine function, and the expansion of the curves P (A) and Q (A) in the Bessel function sequence becomes the Fourier expansion of the amplifier nonlinearity curve y (x) in the sine function sequence. Corresponding to FIG.
Exactly matches the sine waveform of the hysteresis curve y (x) shown in FIG. Theoretically, this type of system should be able to simulate the amplification of a multicarrier signal, that is, a signal composed of multiple sinusoidal components at different frequencies. However, non-linear amplification of a multicarrier input signal is complicated by the occurrence of distortion known as the intermodulation phenomenon. When the nonlinear amplifier receives several carrier frequencies at the input, an undesirable harmonic component, known as a result of intermodulation, is obtained at the output in addition to the amplified carrier, each of which is a carrier component of the carrier. It has a frequency different from the frequency.

FIGS. 11 to 14 show the appearance of an intermodulation phenomenon when nonlinearly amplifying two carrier signals. FIG. 11 is a graph relating to the frequency of the two-carrier input signal x.
They respectively represent two carrier components of frequencies f _-1 and f ₁ , which have the same input amplitude A. The two carrier frequencies f ₋₁ and f ₁ are in the effective frequency band BU of the amplifier and are separated by a frequency difference df.

FIG. 14 shows the two-carrier input signal x of FIG.
7 is a graph relating to the frequency of the output signal y corresponding to the amplified ANLAM with the memory effect of FIG. It can be seen that the output signal y has a series of harmonic components having various frequencies f and different amplitudes C. The frequency of each intermodulation component is an integer combination of the input carrier frequencies. The details of the output harmonics contained in the valid band BU shown in FIG. 14 are as follows:-the frequencies at f _-1 and f ₁ and the amplitude at the output C respectively
₁ and is a _{C 1} carrier; - respectively the amplitude of the output-side frequency at _{f -3} and _{f 3} C
_-3 and C3 _third order intermodulation components.

[0038] Figure 15 reproduces the result of the carrier component and the third-order intermodulation component Abu Elma up tee of model allows the estimation of the amplitude C _-1 and C ₁ and C _-3 and C ₃ of. However, the estimation of the amplitude of the intermodulation component by the Abuermaatti model does not correspond well to the actual measurement of the intermodulation distortion of the nonlinear amplifier in the operation of the multi-carrier with high efficiency.

In fact, when a multicarrier signal is amplified, a phenomenon known as the envelope memory effect occurs. In this case, it cannot be assumed that the envelope is constant as in the case of the model without quasi-memory.
In fact, the time constant of the memory cannot be ignored with respect to the time of the envelope change.

[0040] Figure 12 shows the temporal variation of the second carrier input signal x in FIG. 11, for example, to the amplitude A of each carrier f ₁ and f _-1 have been assumed to be constant, the second carrier signal It shows that the envelopes X (t) and -X (t) are very fast and mostly changing. FIG. 13 showing a temporal change of the output signal y corresponding to the above-mentioned two-carrier input signal is shown in FIG.
This shows that (t) and -Y (t) appear to be greatly distorted by intermodulation distortion. The term “envelope memory effect” is used to refer to such amplitude distortion of a signal of a nonlinear amplifier with memory.

The known models do not particularly focus on the two effects shown in FIGS. The first effect is that the ratio C1 / C3 shown in FIG. 16, that is, the ratio obtained by comparing the carrier amplitude C1 with respect to the amplitude C3 of the intermodulation component, is accurate for the difference df in the carrier frequency of the input signal and the carrier amplitude A. Values vary greatly with different values. When the Abuermaatti model is applied, the ratio of C1 / C3 is independent of the frequency difference and the input amplitude A
It changes continuously according to. The next effect is that the presence of the second carrier affects the output amplitude of the first carrier. Figure 17, for example, depends on the frequency difference between the two carrier _{df (df = f 1 -f -1} ), peak _{C 1} frequency attenuation or resonance of the output amplitude of the carrier is _{f 1} or _{f -1} respectively Or C- ₁ .

One drawback of the known model is that it does not consider such effects. In general, a drawback of known models is that they do not simulate any envelope memory effects.

[0043]

SUMMARY OF THE INVENTION It is an object of the present invention to provide an accurate simulation of the envelope memory effect of a nonlinear amplifier. It is an object of the invention, inter alia, to provide a method and an arrangement which enable an accurate simulation of the response of a nonlinear amplifier to a multicarrier signal. Another object of the present invention is to provide a simple method and apparatus for simulating a nonlinear amplifier with a memory effect.

According to the present invention, these objectives are achieved by performing a dynamic measurement of the characteristics of the amplifier under conditions where distortion occurs. According to the present invention, in each measurement, the amplitude-modulated signal is applied to change the envelope of the signal, and the distortion characteristic of the amplitude modulation is measured. The characteristics of the distortion make it possible to modify the simulated response based on the characteristics plotted at a constant amplitude. In particular, the characteristics are measured by applying a multi-carrier signal, that is, a modulated amplitude.

[0045]

DETAILED DESCRIPTION OF THE INVENTION The invention is achieved by implementing a method for simulating the signal response of a non-linear amplifier exhibiting a memory effect, the method comprising the steps of: Measuring the properties of the amplitude conversion and the amplitude / phase shift conversion, and developing the properties in a sequence of direct transfer functions, the method further comprising:-modulating the input amplitude in each measurement to obtain an amplitude modulation. Calculates the modulation transfer function that reproduces the distortion amplitude at the output side, corresponding to the input modulation amplitude, and directly modulates the input amplitude to simulate the envelope memory effect. Modifying the transfer function.

The method according to the invention preferably comprises the steps of: measuring properties at different frequencies, measuring the frequency correction of the direct transfer function. According to the invention, measuring the characteristic of the amplitude modulation distortion comprises applying a signal comprising at least two different frequency carriers to the input and measuring the intermodulation distortion.

A first method provided by the present invention is to input a group of carriers having a specific amplitude and measure the rejection characteristics of the intermodulation components at the output.
A second alternative provided by the invention is to measure the interaction on the output side with a group of carriers on the input side having a specific amplitude. A third alternative provided by the present invention is to: measure the characteristics of the output modulation noise using a specific input modulation amplitude. According to the invention, preferably: the characteristic of the distortion is measured using the difference of the input modulation frequencies. According to the invention, preferably: the intermodulation distortion characteristic is measured using a plurality of values of the input modulation amplitude.

The invention further provides an apparatus for simulating the signal response of a nonlinear amplifier with a memory effect, the apparatus comprising: a module for calculating the response of the nonlinear amplifier to a single carrier signal; Have at least one transfer function that is derived from the true characteristics of the amplifier, especially the characteristics of the amplitude nonlinearity and the amplitude / phase shift conversion.
With one filter, the transfer function of which is modified in frequency by a modification value calculated based on characteristics measured in a single carrier signal of different frequency, the device generating harmonics responsive to the multi-carrier signal Means.

The harmonic generation means of the device according to the invention may comprise a module for Fourier analysis of the signal into frequency components and / or a coupling module for calculating the frequency combination of the intermodulation components and / or the r.m. m.
Preferably, a standard module for calculating the s value is provided.

In a first embodiment of the device according to the invention, a filter for the modulation of the response of the calculation module, which removes intermodulation components when a multicarrier signal is simulated by the device. It has a transfer function generated from the characteristic.

In a second embodiment of the device according to the invention, a filter for the modification of the response of the calculation module, the filter having a transfer function generated from the interaction properties of the carrier, for which the calculation module The amplitude of a given carrier is modified by a modifying filter when the amplification of the multicarrier signal is simulated by the device.

An advantage of the present invention is that it can be used to simulate the signal response of a microwave amplifier operating at high efficiency. Other features, objects, and advantages of the present invention are set forth in the following detailed description and accompanying drawings, but are not limited to the examples.

[0053]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The simulation method and apparatus according to the present invention have been developed based on a known model system. The invention proposes simulating the amplification of any type of signal, especially a signal containing multiple frequency components. According to a first characteristic of the device according to the invention, the simulation device has means for generating harmonics in response to the multi-carrier signal.

The various structures of the device according to the invention, shown in FIGS. 21 to 23, always include blocks 11, 21 and 31 for the calculation of the output signal y according to the known model and the modification of the output signal y. For this purpose, parallel blocks 12 and 22, or 23 and 32 for generating harmonics are provided for calculating the frequency of the intermodulation component before calculating the exact amplitude. The calculation blocks 11, 21 and 31 for the output signals are known simulation models, for example H.264. B.
Non-linear model MNLSM without memory by Poser,
A. A. M. The second order nonlinear model MNLSM, M.M. T. It is designed to implement each model MM with memory by Abu El Maatti. Mainly M. T. Other models can be implemented based on the improved model of Abu El Maatti, and the description of the implementation of the present invention is developed.

It will be appreciated that a parallel structure of the output calculation block MNLSM or MM and the harmonics generation block Harm is given as an example. Although these blocks schematically represent the subprogram function of calculation in the present invention, the structure of the blocks may be continuous, and a computer program is preferably used to efficiently form calculation blocks. Is good.

FIG. 23 is a detailed diagram of another embodiment of the improved harmonics generating means Harm and the means MM for calculating the intermodulation component based on the simple model of Abuerma Atti shown in FIG. It is a drawing. When the input signal x includes multiple frequency components of A, B,..., M, Y ₀₁₀ , Y _1-11 , Y
_-120 , Y- ₁₁₁ , _Y0-12 , Y- ₁₀₂ , Y
_{20-1, Y 2-10, Y 11-1,} Y 02-1, enabled the calculation of the amplitude of the intermodulation components such. Har
These complex structures of the means m and MM were designed to simulate the amplification of any type of signal, especially multicarrier signals.

In FIG. 20, the harmonics generating means Harm
Is provided with a filter for FFT Fourier series analysis for analyzing the multi-frequency input signal x into frequency components having amplitudes of A, B,..., M. These components are applied to the input of an elementary filter grouped in a table corresponding to the calculation of the intermodulation components, for example _Y010 . Each elementary filter has a transfer function derived from the Bessel function J generalized in the filter example of FIG. Details of the transfer function of the filter table and the generalized Bessel function are described in detail below.

It will be understood that the structure of the device according to the invention preferably comprises means for generating harmonics, preferably means comprising a Fourier series analysis module or a known simulation model.

The expansion of the equations proposed by Abuermaatti's model, which is the most improved of the known models, causes simulation errors as described below. These errors are corrected by the present invention. In fact, following the above equations (7 ') and (7 "), Abuel Maatti separately calculates in-phase nonlinear characteristic P and quadrature nonlinear characteristic Q by the following equations (8') and (8"). Analyzed into a sequence of N Bessel functions:

[0060]

(Equation 10) Here, 2D is the dynamic range of the input of the amplifier, and α _np and α _nq are coefficients calculated by the minimum mean square error method shown in the cited document.

FIG. 10 shows a simulation apparatus of Abuerma Atti corresponding to this analysis. Here, the input signal x is the frequency when the amplitude equal f _-1 and f ₁ is assumed to have two carriers to be A, the signal x is the equation such as Equation (9):

[0062]

[Equation 11]

The input signal x is calculated as shown in FIG.
With envelope X greatly modulated in amplitude as shown in 0):

[0064]

(Equation 12) Here, df = f ₁ −f ₋₁ is equal to two carriers f ₁ and f
Represents a frequency difference of _-1 .

According to the quadratic model without quasi-memory, the envelope Y of the output signal corresponding to the input signal x in equations (9) and (10) is a quadratic equation as follows:

[0066]

(Equation 13) Here, Y _p is the in-phase component of the envelope Y of the output signal y corresponding to the input signal x, Y _q is the quadrature component of the envelope Y.

Here, the equations (11) and (11 ') cannot be used directly for the simulation of the nonlinear amplifier ANLAM with memory. This is because the output envelope X is a sine wave whose frequency is half the frequency difference df, and the change with respect to time is too rapid compared to the time constant of the memory. In the improved model of Abu El Maatti, the envelope Y of the output signal corresponding to the two-carrier input signal x in equations (9) and (10) has a frequency f _i and an amplitude C _i.
(I is a positive or negative integer).

[0068] As shown in detail in Figure 14, the frequency component of the output signal is the frequency and amplitude of the following: - a carrier component of the frequency f ₁ having an output amplitude C _1; - the frequency with output amplitude C _-1 f- ₁ carrier frequencies;-frequencies with unsigned amplitudes of 2. f
_{-1, f -1 + f 1,} 2. a second intermodulation component equal to f ₁ and f ₋₁ −f ₁ ; equal _{f 3} to f ₁ -f _-1, amplitude C
_{A third} intermodulation component that is 3; In f _-1 -f ₁ equal _{f -3,} third intermodulation component amplitude is _{C -3;} - frequency 3. f ₁ -2. In f ₅ equal to f _-1, fifth intermodulation component amplitude is C _5; - components of those and other even-numbered value smaller than the component of the other odd-numbered generally effective band BU Removed from

Only the third and possibly fifth order intermodulation components are located in the effective band BU and have amplitudes C _-3 , C ₃ , C _-5 and C ₅ sufficient for testing the response of the amplifier. . It will be appreciated that the order of the intermodulation component is the sum of the absolute values of the integral coefficients of the carrier frequency combinations.
The corresponding output signal y is the sum of the amplitudes of the frequency components, as in equation (12):

[0070]

(Equation 14) Where i takes an integer value of all of the positive and negative odd, C _i is the real part C _{ip (in} phase) and imaginary part C as shown in Equation
analyzed to _iq (quadrature):

[0071]

(Equation 15)

[0072] In Abu Elmer up Tay supra, the sum of the various Bessel functions Jn is where n next first type as the calculation of the amplitude C _i intermodulation component appearing by amplification of second carrier signals formula We have shown that we can do this based on:

[0073]

(Equation 16) Here, α _np and α _nq are coefficients obtained from the equations (8 ′) and (8 ″) by expanding the nonlinear characteristics P and Q of only the first Bessel function J1, and J0, J1, and J2 are These are the first kind of 0th-order, 1st-order, and 2nd-order Bessel functions.

The 0th-order, 1st-order, 2nd-order, or Nth-order Bessel function J0, J1, J2 or JN is the carrier frequency f when the frequency of the intermodulation component whose amplitude is C _i is f _i.
It is selected to correspond to the absolute value of the integer coefficients of coupling of _-1 or f _1. Then, the | i | next intermodulation component C
_i corresponds to the zero-order, first-order, second-order or N-order Bessel functions J0, J1, J2 or the sum of JN included in each equation. In general, if the input signal x is A, B,. . , M, we have a multiplicity of intermodulation products with frequencies given by the following equation:

[0075]

(Equation 17) Here, α, β,..., Μ are relative integers, and f _A ,
f _B ,. . . , F _M are the carriers A, B,. . , M.

[0076] amplitude C _i of each intermodulation component frequency is at f _i are as the following general formula:

[0077]

(Equation 18) Here, J | n | is the first kind of Bessel function of the order corresponding to the absolute value n (generalization).

FIG. 20 shows a detailed embodiment of the means MM for calculating the intermodulation component obtained by expanding these equations of the model with the memory. This figure shows the components A, B,. . , M is a schematic diagram in which tables of basic filters are connected in series horizontally to calculate a product of a general Bessel function J | α | modulated by the amplitude of M. FIG.
Each horizontal result is then added to another horizontal result in the table or a similar table of quadrature components before giving the harmonic amplitudes Yα, β,..., Μ at the output.

In this detailed embodiment, based on the center carrier B and the carriers A and M at both ends, there are at least nine intermodulation products Y _20-1 , Y except for the amplified carrier Y ₀₁₀ of frequency f _{B. 2-10} , Y _11-1 , Y
_02-1 , _Y1-11 , Y- ₁₂₀ , Y- ₁₁₁ , Y
_0-12, it can be seen _{that Y -102} are obtained in the immediate vicinity of the center frequency _{f B.} Therefore, the number of intermodulation products is
It can be seen that performing exponentially with the number m of carriers, the simulation of the signal of more than one carrier cannot be shown in detail here. One skilled in the art can diagram by generalizing the calculations described for the two carriers.

[0080] The quality determination factors of the design and selection of the microwave amplification system is the ratio of the amplitude C ₁ on the output side of the carrier component of the amplitude C ₃ of the largest intermodulation component of the output side. This factor, referred to herein as the intermodulation rejection factor, is represented by equation (18) below:

[0081]

(Equation 19)

As the P / I ratio increases, the carrier C1 becomes C3
And the quality of this system is improved. In the field of microwaves, the rejection factor is generally represented by current. This corresponds to the following ^proportions: C1 ² / C3 2 here, P / I
Call it ² . In quadratic form, the intermodulation rejection factor can be expressed as:

[0083]

(Equation 20)

The above equation (1) by Abuermaatti
Using the results of 4 '), (14 ") and (16'), (16"), it can be seen that the simulated intermodulation rejection factor for the microwave amplifier hardly changes with frequency. In fact, the frequency correction G _n (f _i ) has a low correction effect, and the curve of the correction G _n is calculated from a single carrier measurement, ie a measurement without modulation of the input envelope. The intermodulation rejection factor P / I thus calculated
Is not a function of the carrier frequency difference df, as opposed to, for example, the resonance effect shown in FIG.

According to a first embodiment of the present invention, the characteristics of the intermodulation rejection are accurately measured and the measured values are used to modify the output signal y calculated based on a known model. Has become. In the following development, the signal y (t) calculated by Abuermaatti's model is
The signal z (t) containing the actually measured intermodulation component is compared. By such a comparison, the value of the modulation signal u (t) is obtained and simulated by the modulation filter according to the invention. FIG. 21 schematically illustrates a first embodiment of the device according to the invention with such a filter 14. Filter 14 has a transfer function E (f, H) that calculates signal u as a function of the input signal. The signal y of the known model 11 is then modified by the modulation signal u to derive a signal z that reproduces the actually measured modulation distortion.

The following development makes it possible to set one example of a relation that can be used to construct the transfer function E (f, H) of the filter IMF for the calculation of the intermodulation. First, it is recognized that the modulated signal u is useful only if the input signal has an amplitude modulated envelope X. In fact, when the input signal x is not modulated, the known model 11 has an output signal y that appropriately simulates the response of the amplifier. According to the present invention, the signal envelope can be obtained by taking the square of the norm of the signal x. Equation (10), which gives an envelope X of a two-carrier signal having an amplitude A and a frequency difference df, is expressed by the following equation (20).
Riding, ^{X 2} is obtained:

[0087]

(Equation 21)

The square equation X ² given by equation (20) can be divided into a constant amplitude A ² and a modulation factor as in equation (20 ′):

[0089]

(Equation 22)

FIG. 21 shows that when the filter 14 is on the input side, the r. m. s. Indicates that it receives the value X ² (t). According to the invention, it is assumed that the modulation signal u coming from the filter 14 can represent the following modified equation:

[0091]

(Equation 23) Where Z is the envelope of the actually measured intermodulated output signal z, Y is the envelope of the signal y calculated with a known model, and U is the envelope of the signal u.

As a result, when there is no modulation at the input, the signal z simulated according to the invention has an envelope Z corresponding to the envelope Y of the simulated output signal y in the known model MNLSM or MM. Have
The modulation signal u is preferably zero. Assume that the envelope U of the modulation signal is the frequency f of the harmonics h on the input side.
And transfer function E (f, H) depending on the amplitude H and r.
m. s. As a result of the value X ² conversion, expressed by the equation (20 '), the such equation of the formula (22):

[0093]

(Equation 24) Here, E (df, A) is a response of a transfer function to a signal having a frequency df and an amplitude A,
(0, A) is a response of a transfer function having a frequency difference of zero.

In general, if there is no difference between the carrier frequencies, it is not necessary to have the modulation signal u, so that E (0, A) is a zero value. K is used as a reference for the following variables: K = A ² . E (df, A) The envelope U of the modulated signal given in equation (22) is preferably written in complex exponential form as follows:

[0095]

(Equation 25)

Here, it is preferable to represent the envelope of the output signal y of the model known by the following complex exponential equation, as in equation (12):

[0097]

(Equation 26) Here, i is an odd integer taking all positive and negative values.

[0098]

[Equation 27]

From equations (21), (23) and (24 '), the following expansion is derived:

[0100]

(Equation 28)

In the first method, it is required to modulate only the amplitude of the third-order intermodulation component by making the minimum change of the first-order carrier component calculated by the known model. This minimizes the value of K and reduces the term K. C- ₃ , K. C _1, K. C- ₁ , K.C. It can be obtained by way negligible the C _3. In this case, the two middle rows of equation (25) indicate that in the modulated output signal z, the carrier components at frequencies f _-1 and f ₁ respectively have amplitudes substantially equal to C _-1 and C _1. Is shown.

First, the amplitude C _{i of the} intermodulation component is
The frequency correction value G shown in equations (12 ') to (25 ")
By eliminating _{np, q} (f _i ), the frequency f _i
Can be assumed to be independent of Therefore, symmetrical components at frequencies _{f -1} and _{f 1} around the carrier _{f -1} and _{f 1} is to have equal amplitudes, such as the identity of the following calculation is simplified: _C -i = _{C i} each end of the second line frequency components _{f 1} and f of formula (23)
₃ , which allow the calculation of the elimination factor P / I, which ultimately has the following values:

[0103]

(Equation 29)

Numerically speaking, the amplitudes C ₁ (A), C ₃ (A) and C ₅ (A) of the components are determined by a known method, and the removal factor P / I whose carrier amplitude is A is determined. After measuring the value of, equation (26) is solved to determine the value of the variable K. For the solution, the variable K is preferably a complex expression such as: K = | K |. exp (j · θ) The argument θ is varied to calculate the value of the norm | K | as a function of the argument θ. The minimization of the changes of the amplitudes C _-1 and C ₁ of the carriers f _-1 and f ₁ , which have been obtained above, is performed by using the argument θ, so that the minimum value of the norm | K | can be obtained. became.

FIG. 18 is a schematic diagram of a measurement system implemented according to the present invention. Difference between carrier amplitudes A, A ', A "and carrier frequency df' = f _-1
Multicarrier signal f to obtain a -f ₁ both change
_-1 and f ₁ series of see that characteristics are determined by performing measurement using. Further, the carrier f ₁ or f ₋₁ used as a reference for the frequency difference df.
It is also possible to obtain a basic change in the frequency f.

As for the frequency difference df, it is possible to measure the amplitude A of the input carrier and the characteristic curve P / I as a function of the current as shown in FIG. Applying the above calculation, a curve of the variable K value as a function of the carrier amplitude A is obtained for each frequency difference df. By obtaining the change in the carrier frequency differences df, df ', df ", df", df "", a table of the curve of the variable K values is obtained. The frame of the transfer function E (df, H) of the filter 14 is set by the table of the curve of the variable K value.

[0107] transmission by the current X ² envelope amplitude H or an input signal of the input side harmonics function E (d
The continuity of (f, H) is obtained by interpolating the discrete values of a table of curves for the variable K. Calculation and interpolation are also performed for all frequency differences df, but the value of the variable K is changed to the selected value df ', df ", df", df "".
It is also possible to interpolate with a frequency difference df between. A computer program performs this kind of calculation,
Preferably, the value of the variable K is stored in a matrix format.

In the second stage, the amplitude f _{i of the} intermodulation component is expressed by the frequency f
Suppose that it depends on _i . Therefore, no longer have identity no longer holds between _{C i} and _{C -i.} The asymmetry in frequency of the response of the amplifier is simulated by such a variant of the first method. In this case,
It has been proposed by the present invention that the transfer function E '(df, H) of the filter 14 may be asymmetric. Equation (22) and the assumptions of the above calculations are then modified by the following complex equation:

[0109]

[Equation 30] Here, E "(O, A) takes the value 0 because this filter does not transmit any continuous component of the signal, and E '(-
df, A) is the response to the frequency difference −df of the transfer function E ′, which is the response E ′ (df, df) to the frequency difference df because the filter E ′ is asymmetric.
Different from A).

The term of equation (22 ') is represented by the following simplified equation: K ⁺ = A ² . E ′ (df, A) K ⁻ = A ² . E ′ (− df, A) Equation (23) is now modified by:

[0111]

(Equation 31)

Let the modulated component of the output signal z 'obtained by such a modification of the first method be C _i '. The output signal z 'is then obtained by modulating the signal y of a known model with the modulation signal u' calculated by the transfer function E 'of the filter 34 in FIG. For this reason,
A distinction can be made between the component C _i calculated in the model MNLSM without memory known as the modulated component C _i ′ according to the invention. Equations (21), (24 '), (2)
According to 8), the modulated component has an amplitude _C -3 as determined by the following formula _{', C -1', C 1} ', C 3':

[0113]

(Equation 32)

Or [0114] Here, removal factor is, intermodulation frequencies either measured when f ₁ and f _3, intermodulation frequency is measured when the f _-1 and f _-3 at a lower carrier at higher carrier Yields two more complex results of intermodulation rejection factors called P / I ⁺ and P / I ⁻ . The two elimination factors have the form:

[0115]

(Equation 33)

Therefore, two variables K are calculated in a known manner.
Solve two equations (33) and (34) using ⁻ and K ⁺ .
When the amplifier has an asymmetric response around the carrier frequency, in order to simulate the modulation distortion at the output side according to the invention, the carrier amplitude A and two rejection factors P / I ⁺ above and below the frequency difference df. And P / I ⁻ measurements to measure the intermodulation rejection characteristics.

The quadratic method according to the invention is used to simulate the quadratic effect of intermodulation distortion. The second order effect is the change in carrier gain as a function of the envelope frequency df / 2 illustrated in FIG. The output amplitude C in the presence of another carrier f- ₁
By measuring the gain of ₁ of the carrier f ₁ of the amplifier, its gain is seen that may change significantly as a function of the difference df between two frequencies of the carrier f _-1 and f _1.

According to the invention, the carrier output amplitudes C ₁ and C _-1 calculated according to the known model can be adjusted taking into account the measurement of the carrier interaction characteristic C 1 / C _-1. It is now possible. The adjustment in the second order method is preferably combined with the adjustment of the main intermodulation components C3 and C-3 planned in the first method of the invention described above. Further, second-order intermodulation components C5 and C5
It is also possible to adjust −5.

In a second embodiment of the device according to the invention, the adjustment modifies the output signal y, or preferably the modulated signal z or z ', using a correction signal v coming from a filter 37 having a correction transfer function F. This is done by: The transfer function F is expressed by the equations (29) to (32) similarly to the function E.
Is calculated by tentatively modifying the modulated component C _i 'formulated above:

[0120]

(Equation 34) Here, L ⁺ = A. F (df, A), L ⁻ = A. F (d
f, A), where F is a modified transfer function expressed as a function of the modulation frequency of the envelope, ie the difference df in the frequency of the carrier amplitude A on the input side.

Components C ₃ ″, C ₋₃ ″, C ₁ ″, C ₋₁ ″
Is called the component to be corrected. Equations (35)-(38) are the four variables K ⁺ , K ⁻ ,
Form a system consisting of four equations with L ⁺ , L ⁻ . These last two equations (37) and (38) and the correcting variable ^{L +} by solving L ^- is given a value. Third order intermodulation component f
₃ and _{f -3,} by accurately measuring the actual amplitude of the carrier _{f 1} and _{f -1,} in known models 1
Next, the third-order, fifth-order component _{C -5, C -3, C -1} , C 1,
After calculating C ₃ and C ₅ , the modulation transfer function E (df, A)
And the frequency correction function F (df, A) can be calculated.

By measuring the amplitude of all components at the output of this type of amplifier, including both the measurement of the third and fifth order intermodulation rejection components at the output and the measurement of the carrier interaction, the nonlinear memory amplifier An overall simulation of the signal response with distortion due to the envelope memory effect is now possible. However, partial measurements such as the third order intermod rejection measurement C1 / C3 and the carrier interaction measurement C1 / C-1 are also sufficient to characterize the particular effect of intermodulation distortion.

In a quadratic manner, when there are a large number of carriers (although the power spectral density DSP of the current is modeled with white noise), the model can be evaluated by measuring the signal-to-noise ratio. It became so. FIGS. 19 (a) and 19 (b) show graphs of the noise frequency used at the input side of the nonlinear memory amplifier and measured at the output side, respectively, to characterize the nonlinear memory amplifier.

In the present invention, as shown in FIG. 19 (a), the power spectrum density DS which is a constant level NP in the frequency band BW except for the noise-free frequency window BH.
White noise with P is applied. This input signal has a variable amplitude, despite having a constant spectral density. Therefore, the modulation distortion appearing on the output side of the amplifier can be measured. The distortion appears at the output of the amplifier in the form of a carryover of the noise relative to the window BH defined above.

The characterization of the amplifier consists of measuring the amplifier's ability to limit the noise carryover to the window BH by measuring the difference between the output noise level NPR in the frequency band BW and the window BH. The measurement of these characteristics is performed at different noise levels NP and is introduced into the simulator, which reproduces the noise characteristics according to the operating environment. A third method of controlling the evaluation of the model consists of measuring the binary error rate in the amplification of the modulated carrier.

FIG. 24 shows a test bench used to measure the characteristics of a non-linear amplifier DUT with a memory effect according to the present invention. This test bench (operating details are well known to those skilled in the art) comprises, in particular, a unit 10 for the generation of an amplitude-modulated signal, which unit applies, for example, a multi-carrier signal to characterize the amplifier DUT. , Two generators SG1 and S2 with different frequencies
G2. Furthermore, the test bench comprises a noise measuring unit 20 having a windowed white noise generator NG, and preferably a comparator for comparing the generator MOD of the signal modulated by a random binary message with the original signal and the return signal from the amplifier. Rator Com
It comprises a unit 30 for measuring a binary error rate with p.

Finally, FIG. 25 illustrates a computer implemented characterization file for simulating nonlinear amplification according to the method of the present invention.
The program Pro now comprises a statistical property AP, a curve C drawn at a constant amplitude in a known manner,
File 20 for storing measurement results of φ or curves P and Q
3 is used. The program Pro also comprises a series of files 2 storing the results of the measurement of the dynamic characteristic P / I, such as the intermodulation rejection curve C1 / C3 or the interaction curve C1 / C-1.
05, 206, and 207 are used. A series of files 205
, 206 and 207 are, for example, the carrier frequency difference df
Created for each.

The program starts after introducing general data Dat into the simulated amplifier. On the output side, the program Pro directly gives the file 204 the transfer function, in particular the coefficient α of the sequence of Bessel functions. Above all, the program according to the present invention includes a file 208 containing the data of the modulation transfer function E and the correction function F described above.
And 209.

FIG. 26 shows an embodiment of a procedure of a method performed by such a program Pro. The device according to the invention is preferably produced in the form of a computer program having functions corresponding to the blocks schematically illustrated by way of example in FIGS. Other embodiments and features and advantages may be improved by those skilled in the art, particularly within the scope of the invention as defined in the following section.

[Brief description of the drawings]

FIG. 1 shows a response curve of the above-described nonlinear amplifier.

FIG. 2 shows a response curve of the nonlinear amplifier described above.

FIG. 3 shows a timing diagram of the signal response of the memoryless nonlinear amplifier or quasi-memoryless nonlinear amplifier described above.

FIG. 4 shows a timing diagram of the signal response of the memoryless nonlinear amplifier or quasi-memoryless nonlinear amplifier described above.

FIG. 5 shows a conceptual diagram of the frequency of the characteristic measurement according to the known principle described above.

FIG. 6 (a) shows the characteristics of a nonlinear amplifier in the form of an amplitude / amplitude conversion curve and an amplitude / phase shift conversion curve obtained by a known method as described above, and FIG. 2 shows the characteristics of a non-linear amplifier in the form of an amplitude / amplitude conversion curve and an amplitude / phase shift conversion curve obtained in a known manner, as described above.

FIG. 7 shows the characteristics of the in-phase and quadrature-phase nonlinearity curve type nonlinear amplifiers obtained by known equivalent methods described above.

FIG. 8 shows a sequence of Bessel functions for characteristic analysis performed by the known simulation method described above.

FIG. 9 shows a known simulation device as described above.

FIG. 10 shows a known simulation device as described above.

FIG. 11 is a frequency graph showing the amplitude of the multicarrier signal on each side of the input and output of the nonlinear amplifier with memory described above.

FIG. 12 is a timing chart showing the amplitude of the multicarrier signal on the input and output sides of the nonlinear amplifier with memory described above.

FIG. 13 is a timing chart showing the amplitude of the multicarrier signal on the input and output sides of the nonlinear amplifier with memory described above.

FIG. 14 is a graph of frequency showing the amplitude of the multicarrier signal on each side of the input and output of the nonlinear amplifier with memory described above.

FIG. 15 reproduces the simulation result of the amplitude of the output frequency component of the non-linear amplifier based on the carrier amplitude of the two-carrier input signal, which was performed by the known simulation method described above.

FIG. 16 shows the amplitude A on the input side of the carrier for the difference df of the various frequencies of the carrier implemented according to the invention.
Of the intermodulation rejection on the output side related to the carrier component at the input of the non-linear amplifier with memory, expressed as a ratio of C1 / C3 corresponding to the ratio of the amplitude of the carrier output to the amplitude of the intermodulation component as a function of Is shown.

FIG. 17 shows a measurement of the interaction of the output carrier of a non-linear amplifier with memory, carried out according to the invention, in the form of the carrier output amplitude y as a function of the carrier frequency difference.

FIG. 18 is a frequency graph showing a measurement of a characteristic performed in accordance with the present invention.

FIG. 19 (a) shows an example of the power spectral density of a signal used to measure the noise-to-power ratio NPR at the input of the amplifier, and FIG. 19 (b) shows the noise-to-power at the input of the amplifier. 4 shows an example of a power spectral density of a signal used for measuring a ratio NPR.

FIG. 20 shows an embodiment of an apparatus for simulating a non-linear amplifier with memory provided with a harmonics generating means according to the present invention.

FIG. 21 shows a first embodiment of an apparatus for simulating a non-linear amplifier with memory according to the present invention.

FIG. 22 shows a modification of the first embodiment of the device for simulating a nonlinear amplifier with memory according to the present invention.

FIG. 23 shows a second embodiment of an apparatus for simulating a nonlinear amplifier with a memory according to the present invention.

FIG. 24 shows a bench used by the method and apparatus according to the invention for measuring the characteristics of intermodulation distortion.

FIG. 25 shows a conceptual diagram of a characteristic file implemented by a method for simulating the response of a nonlinear amplifier according to the present invention.

FIG. 26 shows a conceptual diagram of a procedure of a method for simulating the response of a nonlinear amplifier according to the present invention.

[Explanation of symbols]

 11, 21, 31 Output signal calculation block 12, 22, 23, 32 Harmonics generation block 14 Modulation filter 10 Amplitude modulation signal generation unit 20 Noise measurement unit 30 Secondary error rate measurement unit

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor François-Rene Chevalier France 92160 Antony Boulevard Colbe 24 (72) Inventor Jean-Michel Dumas France 87,000 Limoges Avigny Garibaldi 76

Claims (17)

[Claims] 1. A method for simulating the signal response of a non-linear amplifier exhibiting a memory effect, comprising: measuring the amplitude / amplitude and amplitude / phase shift characteristics of the amplifier at a constant amplitude at the input, respectively; Developing the characteristic in a series of direct transfer functions that sets the amplitude and output phase shift by applying a signal consisting of at least two frequency components to the input and measuring the modulation distortion characteristic of the amplifier. The characteristic of this modulation distortion is measured with respect to some frequency differences and / or some amplitudes of the frequency components, and each measurement causes the amplitude of the output frequency component to be a function of the frequency differences and the amplitude of the input frequency components. The input frequency component when the input amplitude is modulated to simulate the envelope memory effect. Method comprising the steps of calculating a modulation transfer function to correct the amplitude of the output frequency components of the direct transfer function based characteristic of distortion according to the difference and the amplitude of the frequency. 2. The method according to claim 1, comprising: measuring characteristics at different frequencies; and measuring a frequency correction of the direct transfer function. 3. The method according to claim 1, wherein: measuring a characteristic of removing an intermodulation component appearing at an output of the nonlinear amplifier at a frequency different from a frequency of the input component;
Each measurement gives the intermodulation component as a function of the difference and amplitude of the frequency of the input frequency component; from the measurement of the rejection of the intermodulation component, the amplitude of the output component of the direct transfer function, in particular the third order intermodulation frequency component, Calculating a modulation transfer function that sets the modulation signal whose amplitude is to be modulated according to the difference between the frequencies of the input frequency components and the amplitude. 4. The method as claimed in claim 3, wherein the amplifier is asymmetric in frequency, the two factors of rejection of the intermodulation component are measured for each frequency difference and each amplitude of the input frequency component, A method comprising calculating an asymmetric modulation transfer function having a response to a negative frequency difference different from a response to a frequency difference. 5. The method according to claim 1, further comprising: applying a signal comprising at least two frequency components to an input of the non-linear amplifier to measure a characteristic of the interaction of the frequency components; Each measurement gives the output amplitude of the frequency component as a function of the frequency difference of the frequency component and the input amplitude;-from the measurement of the interaction characteristics of the frequency component, the amplitude or the amplitude of the frequency component directly derived from the transfer function; A method comprising calculating a modified transfer function that sets a modified signal for adjusting the modulation amplitude from the difference between the frequency of the input frequency component and the amplitude. 6. The method according to claim 5, wherein the gain of the carrier varies as a function of the envelope frequency, wherein another carrier is present as a function of the difference of the carrier frequencies, the characteristic of the gain at the carrier amplitude. A method that is configured to measure. 7. The method as claimed in claim 1, further comprising the step of: measuring a noise characteristic of the modulation on the output side at a predetermined amplitude of the input modulation. 8. The method according to claim 1, further comprising the step of: measuring a characteristic of the carryover of the noise in the frequency window by inputting white noise having a window. 9. The method according to claim 1, wherein the method is used to simulate the signal response of a microwave amplifier operating with high efficiency. 10. An apparatus for simulating the signal response of a nonlinear amplifier with a memory effect, comprising: a module for calculating the response of the nonlinear amplifier to a single carrier signal; A single carrier signal having two direct filters, the filters having transfer functions generated from the characteristics of the amplitude / amplitude and amplitude / phase shift conversion actually measured in the amplifier, the direct transfer functions taking different amplitudes Wherein the apparatus comprises: a harmonics generating means responsive to the multi-carrier signal; and at least one filter for modulation of the response of the computation module, the modulation filter being the amplitude modulation actually measured in the amplifier. It has a transfer function generated from the characteristics of the distortion, Device transfer function are set in a multi-carrier signal. 11. The apparatus according to claim 10, wherein the module comprises: at least one frequency corrector for directly weighting the transfer function of the filter, each frequency corrector being a different single carrier in the amplifier. An apparatus having a correction function generated from a conversion characteristic measured at a signal frequency. 12. Apparatus according to claim 10 or 11, wherein the harmonics generating means comprises a module for performing a Fourier analysis of the signal into frequency components. 13. The apparatus according to claim 10, wherein the harmonics generating means comprises a coupling module for calculating a frequency coupling of the intermodulation components. 14. The apparatus according to claim 10, wherein the harmonics generating means outputs the signal amplitude r. m. An apparatus comprising a coupling module for calculating an s value. 15. The apparatus according to claim 10, wherein the conversion filter has a transfer function generated from an intermodulation component removal characteristic, and the amplification of the multicarrier signal is simulated by the apparatus. A device in which the amplitude intermodulation component calculated by the modulation filter is added to the response of the calculation module when it is executed. 16. The apparatus according to claim 10, further comprising a filter for modifying a response of the calculation module, the filter having a transfer function generated from an interaction characteristic of the carrier. And an apparatus wherein the amplitude of the carrier provided by the calculation module is modified by a modification filter when the amplitude of the multi-carrier signal is simulated by said apparatus. 17. The apparatus according to claim 10, wherein the apparatus is used for simulating a signal response of a microwave amplifier that operates with high efficiency.